KR102463077B1 - An artificial intelligence smart coaching system and method for coaching various and useful content to users - Google Patents

Abstract

The present invention relates to an artificial intelligence smart coaching system for coaching various and useful content for users, and a method thereof. To this end, the artificial intelligence smart coaching system for coaching various and useful content for users comprises: a user terminal for transmitting, to a smart coaching server, structured data including learning data, evaluation data, and caliper data on students, and unstructured data including consultation data, chatbot data, image data, and log data; and the smart coaching server for providing smart coaching to a user based on the structured data and the unstructured data. Therefore, the learning ability of the user can be effectively improved.

Description

AN ARTIFICIAL INTELLIGENCE SMART COACHING SYSTEM AND METHOD FOR COACHING VARIOUS AND USEFUL CONTENT TO USERS

The present invention relates to an artificial intelligence smart coaching system and method for coaching various and useful content to users.

The development of network technology and the spread of terminals are changing the existing offline-centered learning environment into an online-centered learning environment.

As such, educational institutions that provide online educational content provide related information about educational content, such as basic information and attributes of educational content, to help users select educational content when providing online educational content.

However, even if relevant information on the educational content is provided, there is a limit in selecting the customized educational content suitable for the user's learning level and learning goal. The reason for this is that the user is not sufficiently aware of his/her own learning ability, or that information related to the educational content is not provided in a segmented manner according to the user's learning goal.

The prior art (Korean Patent Registration No. 10-2213476) discloses a learning content recommendation system based on artificial intelligence learning and an operating method thereof.

However, in the prior art, the artificial neural network is trained by assigning weights to the forward sequence and the backward sequence according to the degree of influence on the probability of correct answer based on the problem solved by the user and the response to the problem only, and the artificial neural network is trained with higher accuracy. It only discloses the contents of predicting the probability of the user's correct answer for a specific problem, and does not provide smart coaching for the user with learning recommendations for each level, weak learning recommendations, and special learning recommendations.

Therefore, it is necessary to generate an artificial intelligence model through data received from the user, and provide smart coaching for learning recommendations by level, weak learning recommendations, and special learning recommendations to users based on the generated artificial intelligence model with artificial intelligence. do.

Korean Patent Publication No. 10-2213476

Conventionally, a service for a learning recommendation by level, a weak learning recommendation, and a special learning recommendation is not provided to the user.

Accordingly, the present invention is to provide an artificial intelligence smart coaching system and method for coaching various and useful content to users.

The objects of the present invention are not limited to the above-mentioned objects, and other objects and advantages of the present invention not mentioned may be understood by the following description, and will be more clearly understood by the embodiments of the present invention. It will also be readily apparent that the objects and advantages of the present invention may be realized by the means and combinations thereof indicated in the appended claims.

In order to achieve this object, the present invention provides an artificial intelligence smart coaching system for coaching various and useful contents to users, structured data including learning data, evaluation data, and caliper data for students, counseling data, chatbot data , a user terminal that transmits unstructured data including image data, and log data to a smart coaching server, and the smart coaching server that provides smart coaching to the user based on the structured data and the unstructured data. .

In addition, the smart coaching server of the present invention processes the structured data and unstructured data received from the user terminal, and uses the processed data to create an artificial intelligence model through data labeling, data type change, data combination, data pivot, and data grouping. An artificial intelligence model generation module to generate, an artificial intelligence model maintenance module for updating the artificial intelligence model by tuning the parameters of the generated artificial intelligence model, and a learning recommendation for each level of the student using the updated artificial intelligence model; It may include an artificial intelligence smart coaching module that transmits information on weak learning recommendations and special learning recommendations to the user terminal.

In addition, the present invention provides a method of an artificial intelligence smart coaching system for coaching various and useful contents to a user, wherein the user terminal includes structured data including learning data, evaluation data, and caliper data for students, consultation data, chatbot Transmitting unstructured data including data, image data, and log data to a smart coaching server; and providing, by the smart coaching server, smart coaching to the user based on the structured data and the unstructured data.

In addition, the process of providing smart coaching in the present invention is a process of processing the structured data and the unstructured data received from the user terminal, data labeling of the processed data, data type change, data combination, data pivot, and The process of generating an artificial intelligence model through data grouping, the process of updating the artificial intelligence model by tuning the parameters of the generated artificial intelligence model, and the learning recommendation for each level of the student using the updated artificial intelligence model, vulnerability and transmitting information on the learning recommendation and the special learning recommendation to the user terminal.

According to the present invention, by providing smart coaching to the user based on the structured data and the unstructured data received from the user terminal, it is possible to efficiently improve the user's learning ability.

In addition, the present invention processes the structured data and unstructured data received from the user terminal, and generates an artificial intelligence model through data labeling, data type change, data combination, data pivoting, and data grouping of the processed data, and By tuning the parameters of the generated artificial intelligence model and updating the artificial intelligence model, it is possible to generate and manage an advanced artificial intelligence model.

In addition, the present invention transmits information on a student's learning recommendation for each level, weak learning recommendation, and special learning recommendation through an advanced artificial intelligence model to the user terminal, so that the user can be provided with content optimized for himself/herself.

In addition, the present invention collects structured data and unstructured data from a user terminal, and processes the collected structured data and unstructured data by data labeling, data type change, data combining, data pivoting, and data grouping, and processing the processed data By classifying into a training set, a valid set, and a test set, an artificial intelligence model can be generated.

In addition, the present invention tunes the parameters of the generated artificial intelligence model, verifies the tuned parameters, re-tunes the verified parameters, and re-verifies the re-tuned parameters to upgrade the generated artificial intelligence model. can

In addition, the present invention selects some of the parameters, executes simulations for all cases of the selected partial parameters, and among the results of the executed simulations, the difference between actual data and predicted values by the artificial intelligence model By deriving 5 among the AI models with the fewest and analyzing the overfitting and underfitting of the 5 derived AI models, the AI model with the best performance can be updated.

In addition, the present invention generates a mind map for providing useful content to students by periodically learning the updated artificial intelligence model, and based on the generated mind map, the learning recommendation for each level of the student, the weak learning recommendation and By generating the information on the special learning recommendation, content can be provided through various areas such as parents, friends, and doctors as well as teachers for students.

In addition, the present invention uses the updated artificial intelligence model, selects an item on one subject that is most vulnerable to the student's curriculum, transmits it to the user terminal, and analyzes the response received to the transmitted item to analyze the subject. It is determined whether or not the subject is vulnerable, and when it is determined that the subject is vulnerable, the vulnerability result is provided to the mobile terminal of the teacher who guides the student, thereby improving the student's learning effect in three dimensions.

In addition to the above-described effects, the specific effects of the present invention will be described together while describing specific details for carrying out the invention below.

1 is an exemplary diagram illustrating an artificial intelligence smart coaching system for coaching various and useful content to a user according to an embodiment of the present invention.
2 is a block diagram of a smart coaching server 130 according to an embodiment of the present invention.
3 is an exemplary diagram illustrating an operation of an artificial intelligence model generating module according to an embodiment of the present invention.
4 is an exemplary diagram illustrating an operation of an artificial intelligence model maintenance module according to an embodiment of the present invention.
5 is an exemplary diagram illustrating a parameter tuning operation according to an embodiment of the present invention.
6 is an exemplary diagram illustrating an area of artificial intelligence smart coaching according to an embodiment of the present invention.
7 is a flowchart illustrating a process of a service for recommending weak learning according to an embodiment of the present invention.
8 is an exemplary diagram schematically illustrating the process of the final evaluation of the first unit of FIG. 7 ( S714 ).
9 is an exemplary diagram showing a screen for AI recommendation learning displayed on a user terminal according to an embodiment of the present invention.
10 is an exemplary diagram illustrating an example of using an emoticon according to an expression of an emoticon according to an embodiment of the present invention.

The above-described objects, features and advantages will be described below in detail with reference to the accompanying drawings, and accordingly, those skilled in the art to which the present invention pertains will be able to easily implement the technical idea of the present invention. In describing the present invention, if it is determined that a detailed description of a known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings. In the drawings, the same reference numerals are used to indicate the same or similar components.

Although the first, second, etc. are used to describe various elements, these elements are not limited by these terms, of course. These terms are only used to distinguish one component from other components, and unless otherwise stated, the first component may be the second component, of course.

In the following, that an arbitrary component is disposed on the "upper (or lower)" of a component or "top (or below)" of a component means that any component is disposed in contact with the upper surface (or lower surface) of the component. Furthermore, it may mean that other components may be interposed between the component and any component disposed on (or under) the component.

Also, when it is described that a component is “connected”, “coupled” or “connected” to another component, the components may be directly connected or connected to each other, but other components are “interposed” between each component. It is to be understood that “or, each component may be “connected”, “coupled” or “connected” through another component.

Throughout the specification, unless otherwise stated, each element may be singular or plural.

As used herein, the singular expression includes the plural expression unless the context clearly dictates otherwise. In the present application, terms such as "consisting of" or "comprising" should not be construed as necessarily including all of the various components or various steps described in the specification, some of which components or some steps are It should be construed that it may not include, or may further include additional components or steps.

Throughout the specification, when “A and/or B” is used, it means A, B or A and B, unless specifically stated to the contrary, and when “C to D” is used, it means that there is no specific contrary description. Unless otherwise specified, it means that it is greater than or equal to C and less than or equal to D.

Hereinafter, an artificial intelligence smart coaching system and method for coaching various and useful content to a user according to some embodiments of the present invention will be described.

1 is an exemplary diagram illustrating an artificial intelligence smart coaching system for coaching various and useful content to a user according to an embodiment of the present invention.

Referring to FIG. 1 , an artificial intelligence smart coaching system 100 for coaching various and useful content to a user according to an embodiment of the present invention includes a user terminal (eg, a personal computer (PC) 110 ) and a mobile terminal 120 . )) and the server 130 (eg, a smart coaching server). Hereinafter, at least one of the personal computer (PC) 110 and the portable terminal 120 is referred to as a user terminal.

According to an embodiment, the user terminals 110 and 120 include structured data including student learning data, evaluation data, and caliper data, and unstructured data including counseling data, chatbot data, image data, and log data. Data may be transmitted to the smart coaching server 130 .

In addition, the user terminals 110 and 120 transmit commands and data such that the emoticons for guiding smart coaching and information on the student's level-specific learning recommendations, weak learning recommendations, and special learning recommendations are displayed together on the smart terminal. It may be received from the coaching server 130 . In addition, the user terminals 110 and 120 display information and emoticons received from the server on a display unit (not shown), so that various and useful contents can be coached to students.

According to an embodiment, the smart coaching server 130 may provide smart coaching to the user based on the structured data and the unstructured data.

The smart coaching server 130 processes the structured data and the unstructured data received from the user terminals 110 and 120, and uses the processed data for data labeling, data type change, data combination, data pivot, and data grouping. can create artificial intelligence models.

Also, the smart coaching server 130 may update the artificial intelligence model by tuning parameters of the generated artificial intelligence model. In addition, the smart coaching server 130 may transmit information about a student's learning recommendation by level, weak learning recommendation, and special learning recommendation by using the updated artificial intelligence model to the user terminal through the network 140 .

2 is a block diagram of a smart coaching server 130 according to an embodiment of the present invention.

Referring to FIG. 2 , the smart coaching server 130 according to an embodiment of the present invention may include a communication unit 210 , a memory 220 , and a processor 230 .

The configuration of the smart coaching server 130 shown in FIG. 2 is according to an embodiment, and the components of the smart coaching server 130 are not limited to the embodiment shown in FIG. 2 , and some components as needed may be added, changed or deleted.

According to an embodiment, the communication unit 210 may communicate with the user terminals 110 and 120 and the network 140 through a wired or wireless network, and may transmit/receive various signals or data for smart coaching.

The communication unit 210 may include at least one circuit capable of transmitting and receiving at least one signal or information to and from the communication unit (not shown) of the user terminals 110 and 120 through the network 140 .

According to an embodiment, the communication unit 210 includes data transmitted from the user terminals 110 and 120 (eg, structured data including student learning data, evaluation data, and caliper data, counseling data, and chatbot data). , image data, and unstructured data including log data).

In addition, the communication unit 210 includes data for each student from user terminals owned by various students (eg, structured data including student learning data, evaluation data, and caliper data, counseling data, chatbot data). , image data, and unstructured data including log data). In addition, the communication unit 210 may transmit the received data to the processor 230 .

In addition, the communication unit 210 may transmit information on the learning recommendation for each level of the student, the weak learning recommendation, and the special learning recommendation generated by the processor 230 to the user terminals 110 and 120 through the network 140 . .

According to an embodiment, the memory 220 may store information, data, programs, applications, software, commands, etc. necessary for the operation of the server 130 . In addition, the memory 220 may be, for example, a flash memory type, a hard disk type, a multimedia card micro type, or a card type memory (eg, SD). or XD memory), RAM (RAM), and ROM (EEPROM, etc.) may include at least one type of storage medium.

According to an embodiment, the memory 220 stores a smart coaching algorithm for coaching various and useful content to students with artificial intelligence. Such an algorithm may be implemented by the processor 230 . The artificial intelligence is a program that mimics the human brain neural network, and can support a deep learning algorithm that analyzes, recognizes, infers, and judges various data by itself.

In addition, the artificial intelligence is composed of machine learning (deep learning) and element technologies utilizing machine learning. Machine learning is an algorithm technology that categorizes/learns the characteristics of input data by itself, and element technology uses machine learning algorithms such as deep learning to simulate functions such as cognition and judgment of the human brain. It can perform understanding, reasoning/prediction, knowledge expression, and action control.

According to an embodiment, the memory 220 may store various logics, programs, and instructions that provide smart coaching such as a learning recommendation by level, a weak learning recommendation, and a special learning recommendation to the student by the processor 230 .

According to one embodiment, the processor 230 is ASICs (application specific integrated circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays), It may be implemented as at least one physical element among processors, micro-controllers, and microprocessors.

The processor 230 includes data received from the user terminal (eg, structured data including student learning data, evaluation data, and caliper data, counseling data, chatbot data, image data, and unstructured data including log data). data) can be learned through artificial intelligence algorithms to provide students with smart coaching such as learning recommendations by level, weak learning recommendations, and special learning recommendations.

To this end, the processor 230 analyzes, recognizes, infers, and judges the data by itself, and selects or generates content such as a learning recommendation for each level of a student, a weak learning recommendation, and a special learning recommendation.

According to an embodiment, the processor 230 may include an artificial intelligence model generation module 240 , an artificial intelligence model maintenance module 250 , and an artificial intelligence smart coaching module 260 .

According to an embodiment, the artificial intelligence model generation module 240 may process structured data and unstructured data received from the user terminals 110 and 120 . The structured data may include student learning data, evaluation data, and caliper data, and the unstructured data may include counseling data, chatbot data, image data, and log data.

In addition, the artificial intelligence model generation module 240 may generate an artificial intelligence model through data labeling, data type change, data combination, data pivoting, and data grouping of the processed data.

In addition, the AI model generation module 240 collects the structured data and the unstructured data, and processes the collected structured data and unstructured data by data labeling, data type change, data combination, data pivot, and data grouping. can do. In addition, the artificial intelligence model generation module 240 may classify the processed data into a training set, a valid set, and a test set.

According to an embodiment, the AI model maintenance module 250 may update the AI model by tuning parameters of the generated AI model.

According to an embodiment, the AI model maintenance module 250 may tune parameters of the generated AI model and verify the tuned parameters. In addition, the AI model maintenance module 250 may retune the verified parameters and re-verify the retuned parameters to upgrade the AI model.

In addition, the artificial intelligence model maintenance module 250 may select some of the parameters and execute a simulation on all cases of the selected partial parameters. And, the artificial intelligence model maintenance module 250 derives five from among the artificial intelligence models having the smallest difference between the actual data and the predicted value by the artificial intelligence model, among the results of the executed simulation, and the derivation It is possible to update the AI model with the best performance by analyzing the overfitting and underfitting of the five artificial intelligence models. The number of the derived models is variably adjustable.

According to one embodiment, the artificial intelligence smart coaching module 260 uses the updated artificial intelligence model to transmit information on the learning recommendation for each level of the student, the weak learning recommendation, and the special learning recommendation through the communication unit 210 to the user terminal. It can be transmitted to (110, 120).

According to an embodiment, the AI smart coaching module 260 may generate a mind map for providing useful content to students by periodically performing learning of the updated AI model.

In addition, the artificial intelligence smart coaching module 260 generates information on the learning recommendation for each level of the student, the weak learning recommendation, and the special learning recommendation based on the generated mind map, and based on the generated information Thus, the student can be smart coached.

In addition, the AI smart coaching module 260 selects the questions on the one subject most vulnerable to the student's curriculum by using the updated AI model and transmits them to the user terminals 110 and 120. have.

In addition, the artificial intelligence smart coaching module 260 may analyze the response received from the user terminals 110 and 120 to the transmitted question to determine whether the subject is vulnerable. And, when it is determined that the artificial intelligence smart coaching module 260 is vulnerable to the subject, it may provide a weak result to the teacher's portable terminal (or parent's portable terminal) who guides the student.

According to an embodiment, when the AI smart coaching module 260 selects the questions, if the correct answer rate is less than or equal to a certain percentage, the questions are selected in the order of highest correct answer probability, and the correct answer rate exceeds the predetermined percentage. Then, the items may be selected in the order of the lower probability of correct answers.

In addition, the AI smart coaching module 260 may recommend the special learning based on a learning essential basis for the student, an interest basis, and an issue keyword basis.

And, the artificial intelligence smart coaching module 260 is a command to display the emoticons for guiding smart coaching and information on the student's level-specific learning recommendation, weak learning recommendation, and special learning recommendation together on the user terminals 110 and 120, and Data may be transmitted to the user terminals 110 and 120 .

3 is an exemplary diagram illustrating an operation of an artificial intelligence model generating module according to an embodiment of the present invention.

Referring to FIG. 3 , the artificial intelligence model generation module 240 according to an embodiment of the present invention includes a data collection step 310 , a data processing step 320 , an artificial intelligence model generation step 330 , and service provision and A maintenance step 340 may be performed.

According to an embodiment, the data collection step 310 may collect various types of data, both structured and unstructured. The structured data may include learning data about a student, evaluation data, caliper data, and the like. In addition, the unstructured data may include student counseling data, chatbot data, image (question) data, log data, Internet of Things data, and the like.

According to an embodiment, in the data processing step 320 , the data collected in the data collection step 310 may be processed and analyzed to develop an artificial intelligence model, and may be changed into an easy form. This change may include data labeling, data type change, data combining, data pivoting, data grouping, and the like.

According to one embodiment, the artificial intelligence model generation step 330 may include the step of developing an artificial intelligence model through the data processed in the data processing step 320 . The artificial intelligence model generation step 330 may proceed by classifying data into a training set, a validation set, and a test set.

In addition, the artificial intelligence model generation step 330 may generate an optimal model by tuning at least one parameter that is the core of the artificial intelligence model.

According to an embodiment, in the service provision and maintenance step 340 , useful and various contents may be provided to students by utilizing the results of the artificial intelligence model generated in the artificial intelligence model creation step 330 . In the service provision and maintenance step 340, the service is performed, and learning of the artificial intelligence model may be performed through the collected data. For example, the service provision and maintenance step 340 may include re-performing the AI model development a predetermined number of times (eg, 3 times) or more. And, in the service provision and maintenance step 340, an advanced artificial intelligence model through learning may be applied to the service.

As such, the artificial intelligence model according to the present invention can be advanced through continuous learning even after the service is provided to the student, and various services provided to the student through the advanced artificial intelligence model are all processed through the artificial intelligence algorithm. can

4 is an exemplary diagram illustrating an operation of an artificial intelligence model maintenance module according to an embodiment of the present invention.

Referring to FIG. 4 , the AI model maintenance module 250 according to an embodiment of the present invention includes a data collection step 410 , a data learning step 420 , an artificial intelligence model generation step 430 , and a parameter tuning step. 440 , and a service application step 450 .

According to an embodiment, the data collecting step 410 may include collecting new data after the service is provided to the student.

According to an embodiment, the data learning step 420 may include executing learning through a previously developed algorithm.

According to an embodiment, the step of generating the artificial intelligence model 430 may include generating and verifying the artificial intelligence model based on the learned result.

According to an embodiment, the parameter tuning step 440 may upgrade the model by tuning at least one parameter of the verified artificial intelligence model. In addition, the parameter tuning step 440 may generate an optimal artificial intelligence model (ie, an advanced artificial intelligence model) by repeatedly performing tuning, verification, tuning, and verification.

Such parameter tuning is a unique skill of an analyst, and may require know-how and subjective judgment. And, parameter tuning is to simulate the number of at least one parameter case that can be set in an artificial intelligence model and select a model with the smallest error among them. In addition, this selection fully considers overfitting and underfitting. These parameter procedures will be described later.

According to an embodiment, the service application step 450 may apply the advanced artificial intelligence model to the actual service.

As such, the AI model maintenance module 250 can upgrade the model through re-learning when new data is collected, and this process is not finished once, but is performed regularly and continuously while the model is being serviced. can be

And, the artificial intelligence model developed through the artificial intelligence algorithm collects new data when the service starts, and when new data is accumulated, re-learning is performed, and the upgrading process is performed through this re-learning. The upgrading process is to increase the completeness of the AI model by tuning at least one parameter. When the advancement is completed, the reliability of AI smart coaching can be improved by applying the new model to the service.

5 is an exemplary diagram illustrating a parameter tuning operation according to an embodiment of the present invention.

Referring to FIG. 5 , the parameter tuning step 440 according to an embodiment of the present invention includes a step 510 for setting the number of parameter cases, a step 520 for simulation, and a step 530 for considering overfitting and underfitting. ), and a model selection step 540 .

In creating an artificial intelligence model, there are various parameters that can be set. Among them, it is the analyst's know-how to set the minimum number of possible parameters for an AI model with excellent performance, and the less the number of cases, the shorter the simulation time. According to an embodiment, the step of setting the number of parameter cases ( 510 ) may set the number of parameter cases in order to shorten the simulation time.

According to an embodiment, in the simulation step 520 , a simulation is executed for the number of all instances of the parameter set in the step 510 for setting the number of parameter instances. And, the simulation step 520 derives a predetermined artificial intelligence model (eg, five artificial intelligence models) according to the priority among the models having the smallest difference between the actual data and the predicted value by the model among the executed simulation results. do.

According to an embodiment, the step of considering the overfitting and the underfitting ( 530 ) may include selecting a predetermined artificial intelligence model (eg, five artificial intelligence models) from among the simulation results in the simulation step ( 520 ). Analyze overfitting and underfitting of intelligence models.

According to an embodiment, the model selection step 540 selects a model with the best performance through the step 530 considering the overfitting and underfitting. And, the selected model is the best model in the corresponding version, and it continues to evolve through periodic tuning.

6 is an exemplary diagram illustrating an area of artificial intelligence smart coaching according to an embodiment of the present invention.

Referring to FIG. 6 , AI smart coaching according to an embodiment of the present invention may be applied to various fields such as parents, teachers, services, doctors, and friends. For example, a parent may receive services for education, evaluation, counseling and management through artificial intelligence smart coaching according to the present invention, and a student's teacher may receive services for counseling, evaluation, management and education. In addition, the doctor may receive services for emotion diagnosis, blood pressure diagnosis, stress diagnosis, and disease diagnosis, and may provide services for study, play, and counseling to students' friends.

For example, AI smart coaching according to an embodiment may recommend learning by level, recommending weak learning, and recommending special learning when a teacher educates a student. In addition, the AI smart coaching may provide a personalized timetable for each student, and may also provide a real-time question answering service.

[Table 1] below shows an example of content provided through artificial intelligence smart coaching according to an embodiment.

classification service name Contents teacher AI Question Answering System Advancement of real-time question answering service in middle and high school -> Take a picture of a problem you do not know while learning -> Real-time automatic answer
AI algorithms allow you to receive matching or similar answers to questions AI Concept Test Concept tests exist in all lectures in elementary, middle and high schools / Test progress form after watching lectures -> Provide different types of concept tests for each learner through AI algorithm / Customized learning with a good concept lecture method after pre-test AI Civil War After learning the recommended problem in my national rank (seat), I provide a recommended course according to my rank -> Segment the group of students through an AI algorithm and recommend the appropriate study method (course), linked learning, etc. of the group that the user belongs to service AI information box Auto-response function when learners or parents inquire about personal information (the form of use of chatbot)
Example: Card-related information, class time, contract expiration time, end time regarding cups or books you own, etc. AI Alarm Learner management Send VMS so that the management teacher directly sends an alarm on the CRM during class->Automatically alarms before class starts. Voice call and search function In case of ice cream home run, automatic voice call is not possible, voice recognition is possible after accessing the page, there is no connection with learning due to exposure of the nator search result value during voice search -> After the tutor automatically recognizes when the name is called (eg Hello Jennybot) Service available, learning and searching through voice during learning Provision of AI entrance examination/learning information Currently, there is a menu for entrance exam/career, but the content is poor and the use is low -> Provides information by recommending contents related to entrance exam/study among the contents searched by many middle and high school students bookmark If you press the wrong button while learning, you need to find and access the corresponding content to open a pop-up window that can be connected to the last learning point -> A function that can be displayed like a bookmark without entering the content you were learning friend AI Reading Room
(Study Friends) Study service in the form of studying together -> Applied in the reading room rather than all learning, you can also see the current status of other students. You can set it to cover your face with a character, etc. when you are reluctant to expose your face parents AI psychological counseling service Solving difficult problems through non-face-to-face solutions Collect the results as data in connection with counseling through tests such as psychological tests and career tests tailored to students' eye level -> Learning through AI algorithms -> School life, career path, various Providing coaching and solutions to concerns doctor Stress Diagnosis Apply the service currently provided by the smart watch in conjunction with milk tea -> Provide statistical information such as alarm/stress time and day of the week when stress increases while learning

According to an embodiment, the AI smart coaching includes a service that recommends weak learning of milk tea users in addition to the contents of [Table 1] (eg, one bite recommendation service) and learning related to interesting learning and issues other than the regular course It may include artificial intelligence special learning that recommends

7 is a flowchart illustrating a process of a service for recommending weak learning according to an embodiment of the present invention.

Hereinafter, referring to FIG. 7 , in the service for recommending weak learning according to an embodiment of the present invention (eg, one bite recommendation service), when an AI customized course is registered (S710), the student learns the first unit (S712). Then, through the final evaluation of the first unit (S714), it is determined whether the student's learning is weak (S716). If the student's learning is weak, it is determined as unit weak learning (S718). Such weak learning can be judged through basic problems and unit final evaluation.

And, a service that recommends weak learning (eg, a one-bite recommendation service) repeats this process. For example, a service for recommending weak learning (eg, one bite recommendation service) learns the N-th unit ( S720 ). Then, it is determined whether the student's learning is weak through the final evaluation (S722) of the Nth unit (S724). If the student's learning is weak, it is determined as unit weak learning (S726).

A service that recommends vulnerable learning (eg, one bite recommendation service) provides students with content and items for weak learning on one of the most vulnerable subjects, and provides them to students or teachers after final evaluation of the unit.

Whether a student is vulnerable to learning is judged through [Table 2] below.

Understanding the concept (bottom) Basic improvement (medium) Completion of skills (award) the best standard ~60 61-90 91~94 95~

In addition, a service that recommends weak learning (eg, a nibble recommendation service) can determine whether it is vulnerable by performing a MAT evaluation (S728) (S730, S732). A service that recommends weak learning (eg, a nibble recommendation service) ) can be judged weak through basic problems, unit final evaluation, and achievement evaluation. And, after providing the students with the content and questions for the weakest learning for the 1st weakest (two maths) unit, evaluating the achievement, the learning is provided to the student in a specific class the next day.

Whether a student is vulnerable to MAT learning is judged through [Table 3] below.

Understanding the concept (bottom) Basic improvement (medium) Completion of skills (award) the best standard ~60 61-90 91~94 95~

8 is an exemplary diagram schematically illustrating the process of the final evaluation (S714) of the first unit of FIG. 7. Referring to FIG. 8, when a weak recommendation service is provided by linking an artificial intelligence algorithm according to an embodiment of the present invention , the first unit final evaluation (S714) checks the target topic to be determined, determines the weak topic as a result of the determination, and learns the weak topic.

In the step of confirming the topic to be judged, it is determined whether the subject is vulnerable by the item history value and the predicted value. And, the greater the difference between the history result and the prediction result in each item topic, the higher the vulnerability ranking. For example, if the history result of a specific item topic (eg, Sc0301_010201) is 50 and the prediction result is 70, the vulnerability rank is 1. And, if the history result of another specific item topic (eg, Sc0301_010101) is 90 and the prediction result is 80, the vulnerability rank is 3.

And, in the weak topic learning step, there are a weak content M code and a weak question M code for each item topic, and includes a weak question (ie, incorrect answer, correct answer, new according to the quiz code).

For example, the weights for calculating the percentage of correct answers for each topic are shown in [Table 4] below.

basic problem end-of-unit problem total Correct answer history (correct answer rate) 30% Probability of correct answer 10% 20% 30% total 60% 100%

And, for example, the number of targets for providing vulnerable re-learning is shown in [Table 5] below.

Korean english math Social science Unit final evaluation One One One One One performance evaluation One One 2 One One Target grade Grades 1-6 Grades 3-6 Grades 1-6 Grades 3-6 Grades 3-6 note D+1, D+21 provided

In addition, for example, the logic of providing a learning weak item is shown in [Table 6] below.

division correct answer rate number of incorrect answers division number 1 No.2 number 3 Case 1 less than 50% 1 or less noon answer wrong answer new question new question Probability of correct answer IRT difficulty 2 (initial setting) 2 (initial setting) Case 2 less than 50% 2 noon answer wrong answer wrong answer new question Probability of correct answer highest 2nd highest IRT difficulty 2 (initial setting) Case 3 less than 50% 3 or more noon answer wrong answer wrong answer answer question Probability of correct answer highest 2nd highest lowest IRT difficulty Case 4 50% or more
~75% less 1 or less noon answer wrong answer wrong answer new question Probability of correct answer lowest IRT difficulty 2 (initial setting) Case 5 50% or more
~75% less 2 noon answer wrong answer wrong answer answer question Probability of correct answer the highest problem 2nd highest problem lowest problem IRT difficulty Case 6 50% or more
~75% less 3 or more noon answer wrong answer wrong answer answer question Probability of correct answer the highest problem 2nd highest problem 3rd highest problem IRT difficulty Case 7 75% or more 1 or less noon answer wrong answer correct answer question answer question Probability of correct answer lowest problem 2nd lower problem IRT difficulty the highest problem 2nd highest problem Case 8 75% or more 2 noon answer wrong answer wrong answer answer question Probability of correct answer the highest problem 2nd highest problem lowest problem IRT difficulty Case 9 75% or more 3 or more noon answer wrong answer wrong answer wrong answer Probability of correct answer the highest problem 2nd highest problem 3rd highest problem IRT difficulty

According to an embodiment, the AI smart coaching module 260 selects the questions in the order of the highest probability of correct answer when the correct rate is less than a certain percentage (eg 50%), and the correct answer rate is the predetermined percentage (eg 50%) %) or more, the questions may be selected in the order of the lowest probability of correct answer.

9 is an exemplary diagram showing a screen for AI recommendation learning displayed on a user terminal according to an embodiment of the present invention.

Referring to FIG. 9 , the smart coaching server 130 displays information on the student's level-specific learning recommendation, weak learning recommendation, and special learning recommendation and an emoticon 914 guiding smart coaching together on the user terminals 110 and 120. It is possible to transmit a command and data to make it possible to the user terminals 110 and 120 .

The user terminals 110 and 120 may display a screen 910 for recommended learning based on commands and data received from the smart coaching server 130 on a display unit (not shown).

The screen 910 may include an artificial intelligence learning menu 911 , an artificial function popular menu 912 , an artificial paper issue menu 913 , and an emoticon 914 . For example, the screen 910 may include a message 915 to be delivered to a student through an emoticon 914 (name: Zinia).

The emoticon 914 is a character used in the AI smart coaching described in the present invention, and performs a function of intuitively providing the AI smart coaching service to students.

10 is an exemplary diagram illustrating an example of using an emoticon according to an expression of an emoticon according to an embodiment of the present invention.

Referring to FIG. 10 , the first emoticon 1011 shows a normal expression as the facial expression is expressionless, and the second emoticon 1012 is a facial expression that expresses happiness when the correct answer is correct, when the score is good, etc. used in the situation. In addition, the third emoticon 1013 is used in a situation where the face expression is a wink when the correct answer is correct, when the score is good, etc., and the fourth emoticon 1014 is used when the learning starts and ends as the face expression is a smile. do.

In addition, the fifth emoticon 1015 shows a case in which the facial expression is exciting and the recommended AI recommendation special learning is all studied on the day, and the sixth emoticon 1016 is a face expression and has not yet completed learning. It indicates when to do it. In addition, the seventh emoticon 1017 represents a case in which there is no action for several minutes during learning, such as diagnostic evaluation, as the face expression is angry, and the eighth emoticon 1018 is a sad face expression and has not yet completed learning. It indicates when to do it.

In addition, the ninth emoticon 1019 represents loading that brings results such as diagnostic evaluation as a facial expression loading, and the tenth emoticon 1020 is a simple sentence with a simple facial expression, such as HELLO, HI, GOOD, GREAT, etc. , indicating the expression of the sentence.

In this way, the emoticon 914 of the present invention can express emotions according to various situations, and facial expressions and actions can be changed according to the emotions.

As described above, the present invention can coach a variety of useful content to students (or users) learning through artificial intelligence smart coaching.

Each step in each of the above-described flowcharts may be operated regardless of the illustrated order, or may be performed simultaneously. In addition, at least one component of the present invention and at least one operation performed by the at least one component may be implemented in hardware and/or software.

As described above, the present invention has been described with reference to the illustrated drawings, but the present invention is not limited by the embodiments and drawings disclosed in the present specification. It is obvious that variations can be made. In addition, although the effects according to the configuration of the present invention are not explicitly described and described while describing the embodiments of the present invention, it is natural that the effects predictable by the configuration should also be recognized.

110: personal computer 120: mobile terminal
130: smart coaching server 140: network
210: communication unit 220: memory
230: processor

Claims (10)

In the artificial intelligence smart coaching system that coaches various and useful contents to users,
a user terminal for transmitting structured data including learning data, evaluation data, and caliper data for a student, and unstructured data including counseling data, chatbot data, image data, and log data to the smart coaching server; and
Based on the structured data and the unstructured data, comprising the smart coaching server that provides smart coaching to the user,
The smart coaching server,
Artificial intelligence model generation that processes the structured data and the unstructured data received from the user terminal, and generates an artificial intelligence model through data labeling, data type change, data combination, data pivoting, and data grouping of the processed data module;
an artificial intelligence model maintenance module for updating the artificial intelligence model by tuning the parameters of the generated artificial intelligence model; and
and an artificial intelligence smart coaching module that transmits information on the student's level-specific learning recommendation, weak learning recommendation, and special learning recommendation using the updated AI model to the user terminal,
The artificial intelligence model maintenance module,
A simulation is executed by selecting some of the parameters, and based on the executed simulation, predetermined artificial intelligence models having the smallest difference between actual data and a predicted value by the artificial intelligence model are derived, and the predetermined artificial intelligence model is derived. An artificial intelligence smart coaching system that analyzes overfitting and underfitting of intelligence models to update the best performing AI model.
The method of claim 1,
The artificial intelligence model generation module,
Collect the structured data and the unstructured data, and process the collected structured data and unstructured data by data labeling, data type change, data combining, data pivoting, and data grouping,
An artificial intelligence smart coaching system that classifies the processed data into a training set, a valid set, and a test set.
The method of claim 1,
The artificial intelligence model maintenance module,
Artificial intelligence smart coaching to upgrade the artificial intelligence model by tuning the parameters of the generated artificial intelligence model, verifying the tuned parameters, re-tuning the verified parameters, and re-verifying the re-tuned parameters system.
4. The method of claim 3,
The artificial intelligence model maintenance module,
Five of the artificial intelligence models with the smallest difference between the actual data and the predicted value by the artificial intelligence model are derived,
An artificial intelligence smart coaching system that updates the best performing artificial intelligence model by analyzing the overfitting and underfitting of the five derived artificial intelligence models.
The method of claim 1,
The artificial intelligence smart coaching module,
A mind map is generated to provide useful content to the student by periodically conducting the learning of the updated artificial intelligence model,
Based on the generated mind map, generate information on the learning recommendation for each level of the student, the weak learning recommendation, and the special learning recommendation,
Based on the generated information, an artificial intelligence smart coaching system for smart coaching the student.
The method of claim 1,
The artificial intelligence smart coaching module,
Using the updated artificial intelligence model, selects an item on one subject most vulnerable to the student's curriculum and transmits it to the user terminal,
Analyze the response received for the transmitted question to determine whether the subject is vulnerable,
If it is determined that the subject is vulnerable, an artificial intelligence smart coaching system that provides a weak result to the mobile terminal of the teacher who guides the student.
7. The method of claim 6,
The artificial intelligence smart coaching module,
If the percentage of correct answers is less than a certain percentage, the above items are selected in the order of the higher probability of correct answers,
If the correct rate is greater than or equal to the predetermined rate, an artificial intelligence smart coaching system for selecting the items in the order of the lower probability of correct answers.
7. The method of claim 6,
The artificial intelligence smart coaching module,
An artificial intelligence smart coaching system that recommends the special learning based on a learning essential basis, an interest basis, and an issue keyword basis for the student.
The method of claim 1,
The smart coaching server,
An artificial intelligence smart coaching system that transmits commands and data to the user terminal so that an emoticon for guiding smart coaching and information on the student's learning recommendation for each level, weak learning recommendation, and special learning recommendation is displayed together on the user terminal.
In the method of an artificial intelligence smart coaching system for coaching various and useful content to users,
Transmitting, by the user terminal, structured data including learning data, evaluation data, and caliper data for the student, and unstructured data including counseling data, chatbot data, image data, and log data to the smart coaching server; and
The smart coaching server, based on the structured data and the unstructured data, includes a process of providing smart coaching to the user,
The process of providing the smart coaching is,
The process of processing the structured data and the unstructured data received from the user terminal,
The process of generating an artificial intelligence model through data labeling, data type change, data combining, data pivoting, and data grouping of the processed data;
The process of updating the artificial intelligence model by tuning the parameters of the generated artificial intelligence model, and
using the updated artificial intelligence model to transmit information on the student's level-specific learning recommendation, weak learning recommendation, and special learning recommendation to the user terminal,
The process of updating the AI model is:
The process of running a simulation by selecting some of the parameters;
A process of deriving predetermined artificial intelligence models having the smallest difference between actual data and a predicted value by the artificial intelligence model, based on the executed simulation, and
and updating the AI model with the best performance by analyzing the overfitting and underfitting of the predetermined AI models.

Images